On Thu, Jan 27, 2011 at 01:27:57PM +0100, Marek Lindner wrote:
> 
> Hi,
> 
> > > + shost = wifi_hdr->addr2;
> > > + if (fc & IEEE80211_FCTL_FROMDS)
> > > +         shost = wifi_hdr->addr3;
> > > + else if (fc & IEEE80211_FCTL_TODS)
> > > +         shost = wifi_hdr->addr4;
> > > +
> > > + dhost = wifi_hdr->addr1;
> > > + if (fc & IEEE80211_FCTL_TODS)
> > > +         dhost = wifi_hdr->addr3;
> > 
> > What would happen if you happened to pick up a WDS packet? Four
> > addresses in the packet. Do we see sensible addresses?
> 
> we should - that is what addr4 is used for but I did not test it. Let me know 
> if it does not work for you.
> 
> 
> > > +
> > > + hdr_len = 24;
> > > + if ((fc & IEEE80211_FCTL_FROMDS) && (fc & IEEE80211_FCTL_TODS))
> > > +         hdr_len = 30;
> > 
> > Again, WDS? We are now a few bytes out when decoding the rest of the
> > packet.

OK. Sorry. I'm wrong.

What confused me is time spent debugging WDS in madwifi. It has macros
similar to what you have:

#define IEEE80211_FC1_DIR_NODS                  0x00    /* STA->STA */
#define IEEE80211_FC1_DIR_TODS                  0x01    /* STA->AP  */
#define IEEE80211_FC1_DIR_FROMDS                0x02    /* AP ->STA */
#define IEEE80211_FC1_DIR_DSTODS                0x03    /* AP ->AP  */

So i was expecting to see something like DSTODS. However the madwifi
macros are for the complete DIR field, where as you are looking at
individual bits. So 

(fc & IEEE80211_FCTL_FROMDS) && (fc & IEEE80211_FCTL_TODS) 

is equivalent to the madwifi IEEE80211_FC1_DIR_DSTODS.

So you header length calculation looks O.K.

   Andrew

Reply via email to